V.D.2.g In-Band Seconds-Confirmation Confusion Processing

IRAS Explanatory Supplement
V. Data Reduction
D. Point Source Confirmation
D.2 Overview of Seconds-Confirmation
D.2.g In-Band Seconds-Confirmation Confusion Processing

Chapter Contents | Introduction | Authors | References
Table of Contents | Index | Previous Section | Next Section

The drop-dead detection and all candidates which had passed the fine position test were immediately marked as confused upon initiation of confusion processing. Status bits set here and in corresponding subsequent analyses were carried in a status word, hereafter denoted CSTAT, throughout all remaining processing stages (see Table V.D.3a,b).

Several attempts were made to unravel confused situations. The first stage involved discarding any detections with signal-to-noise ratios below 5 or detection correlation coefficients below 0.92. Such detections were never used in any future confirmation processing. If this dropped the number of candidates to zero or one, then the drop-dead was processed for rejection or for the double-detection mode, respectively. Otherwise confusion processing continued by attempting to identify pairs of candidates which would confirm each other without the help of the drop-dead detection and without confusion from other detections in the buffer. Any such detection pairs were set aside from the current situation and left for the time when one of them would become a drop-dead. Finding such pairs reduced the number of candidates involved in the confused situation being processed, and after all such possible reduction had been achieved, the number of remaining candidates was checked as described above.

If the situation remained confused then processing continued by casting aside the detections whose slots had an edge-overlap relationship with the drop-dead's slot; if any, unless the detector status information indicated that these were allowable for pairwise confirmation with the drop-dead. Again the branching possibilities were checked. In this case, and in all remaining cases in this section, discarded candidates remained eligible for future confirmations on subsequent processing.

Confusion Status (CSTAT) Bit Assignments
Table V.D.3a
Bit No.Meaning (Applies if Bit = 1)
0Either triple-detection mode or possible
confusion in seconds-confirmation
1Triple-detection mode accepted
2Confusion diagnosed and cleanup attempted
in seconds-confirmation
3Confusion diagnosed in band-merging, only
best match retained
4Confusion diagnosed in hours-confirmation,
only best match retained
5In 12 µm Band Byte: Not Used
In 25 µm Band Byte: Not Used
In 60 µm Band Byte: Detector number arrays
(all bands) are out time order
In 100 µm Band Byte: More then 3 sightings were hours-confirmed
6ADC saturation occurred for at least one detection in this band
7Not used

Common CSTAT Values (Decimal)
Table V.D.3b.
3Triple-detection mode
5Possibly confused
7Triple-detection mode after confusion cleanup; usually not reliable
8Possibly confused in band-merging
13Possibly confused in both seconds-confirmation and band-merging
16Possibly confused in hours-confirmation

If further candidate reduction was needed, the candidates were required to pass a stricter position test, similar to the one discussed above, but with a threshold value such that one real detection pair out of one hundred would be expected to fail this test. Again the branches considered above were checked except this time the case of two remaining candidates was included, with a branch to the triple-detection mode processing if applicable, provided that a branch from that mode to confusion processing had not already been executed.

If more than two candidates remained at this point (or two which had been found unacceptable for the triple-detection mode), then that combination whose summed fluxes came closest to those of the drop-dead were retained and the others were discarded. If this left no candidates, the drop-dead was rejected; if only one candidate remained the double-detection mode processing was performed. Otherwise, a flux test between the summed candidate fluxes and the drop-dead flux was performed which was similar to the triple-detection mode flux test, except that the threshold value used was chosen so that one real case out of a million would be rejected. If this test was passed then all remaining candidates were confirmed with the drop-dead which was considered to be a simultaneous confused observation of all of them. This type of salvaging was possible only when the drop-dead was the confused sighting (rather than one of the candidates). In this case, no flux refinement was performed; the individual candidate fluxes were retained, but position refinement was performed as described below.

Chapter Contents | Introduction | Authors | References
Table of Contents | Index | Previous Section | Next Section